Aapo Talvensaari
Aapo Talvensaari
@Water-Melon does it mean we cannot get this `old` working anymore, even with SECLEVEL=0? https://wiki.mozilla.org/Security/Server_Side_TLS#Old_backward_compatibility @fffonion is that right? If so, then we need to remove the whole `old` thing...
> Therefore, this may indicate that openssl `3.x`'s `s_client` no longer supports `TLSv1.x`. But it (3.x) works on server with TLSv1.1/TLSv1.0? Or not?
The non-varargs optimization can be added if neccessary. (Edit: I added small optimization for that on second commit).
> Since we need to profile and benchmark this PR, could you please convert this PR as a draft? Converted to draft.
> Don't forget to remove the first argument at the definition of `run_hooks()` > > https://github.com/Kong/kong/blob/d43159afcec9212a4e214b7568c4922a832c1651/kong/init.lua#L325 I have removed the `ctx` from `run_hooks`.
> Should this PR be released in 3.7? Let's try to get it in. It is not a must, but I see no point of post-poning.
I am removing incomplete-cherry-pick as there is now manual one: https://github.com/Kong/kong-ee/pull/8863
Looks to be same as update: https://github.com/Kong/kong/blob/master/kong/api/routes/upstreams.lua#L216-L221 Not sure, but this may be because of historical purposes. But doc seems be then wrong. One of them is wrong.
I discussed this with @locao, and it looks like it is a bug (in code).
@outsinre why is this reopened. I already said that this defeats the whole purpose of coroutine mutex?